RA_Fisher

Description has to come first, then we can form hypotheses.

Yes, you're referring to transfer learning. It's an empirical question that also depends on the nature of activities.

Oof. It’s so sad when businesses with in-demand products suffer from a lack of successful economic mechanism design.

Europeans use American iPhones and Android bc they wouldn’t add a backdoor.

Didn’t the UK want a backdoor added?

Yes indeed, I call it the revenge of statistics! Pseudoreplication rears its ugly head once again. Nonparametrics are certainly useful, but once you know things about the DGP, it’s often possible to do something “traditional” that’s better. Probabilistic programming, for life!

AI be used to make reviewing it easy, eh?

Exactly, it’s unrepresentative of AI. It’s damaged AI.

We need a new Pelican, bc earlier models were disadvantaged relative to later models. It's not a very useful comparison anymore.

I agree. I don't think there's much public information out there about the due diligence involved in his hiring.

Fair. An interesting question: how quickly can we detect something without being thwarted by anisotropy / the multiplicity of backward paths? ie- retrodiction

Let’s organize the temporal order a bit. This is what some research turned up.

“Groups of senior employees, concerned with Altman’s leadership and lack of transparency, asked Loopt’s board on two occasions to fire him as C.E.O., according to Hagey.”

“As Mark Jacobstein, an older Loopt employee who was asked by investors to act as Altman’s “babysitter,” later told Keach Hagey, for “The Optimist,” a biography of Altman, “There’s a blurring between ‘I think I can maybe accomplish this thing’ and ‘I have already accomplished this thing’ that in its most toxic form leads to Theranos,” Elizabeth Holmes’s fraudulent startup.”

https://ghostarchive.org/archive/veLhW

I read he’s a vegetarian out of concern for animals, and that’s a good moral sign (but perhaps less relevant here).

Are / were there precursor concerns arising to signal back then? I don’t know.

It’s well-known Brockman and his wife donated $25 million to MAGA Inc., the main super PAC supporting Donald Trump.

The main higher-level factor is our patriarchal culture (and more bad things tend to stem from places where that’s intense).

This is super, but students will have access to AI during the test in real life, so it's ironically less realistic to remove it (thinking of the "... GPT-4 actually harmed subsequent performance by 17% when the tool was removed ..." part).

I'm more curious how students perform on the test with vs. without AI.

So Opus might be correct?

An LLM disproved a famous 80-year-old conjecture.

True, during Covid when blue collar in-person maga unemployment skyrocketed, the Fed could have avoided increasing the money supply. I wish they had, bc we all had to pay the resulting inflation cost. How were thanked for it? Tons of Constitutional law-breaking.

There’ll always be boundary tending, true. Only a portion of CS deals with stochastic functions though, whereas all of statistics is stochastic. That makes a big difference, bc the world is complex.

Information theory doesn’t even incorporate utility.

Programming is a lot easier than statistics bc it’s deterministic, whereas statistics is stochastic (that extends and encompasses deterministic functions).

AI speeds up learning, so I bet that’s what you’re noticing with R.

As an aside, the best programmers these days are probabilistic programmers (who write stochastic functions). Our languages are Stan and PyMC. Both can be called by Python or R, and AI writes all of them extremely well. So it seems to me that the underlying language matters less than ever.

Companies should encourage AI use in interviews to avoid this issue.

The gold standard and metalism generally, leads to all kinds of unproductive panics bc the quantity of money can’t wisely be adjusted to the situation. It’s a bad trade off, bc it’s well-known in the literature that inflation-targeting works (and that’s the current world-wide central bank policy since 1991).

I used AI to unpack it a bit here: https://statwonk.com/econometricians-can-build-decision-engi...

I'd generally point to econometrics and statistics applied to business. The key activity is causal inference and then the context determines the mix of econo vs. stats required to help the org make high-quality decisions to increase output or make it more lucrative or higher-quality.

Econometricians can solve it, bc we can create rigorous models that map causal inputs to output.

It’s extremely advanced technology, though, and most CEOs would rather rent seek / camp than give up some decision-making power (and very few are even aware it’s possible).

It may be that they’re protecting their time.

True, but high-quality education, whether personal or formally, tends to produce high-quality knowledge.

Yes, companies pay for what’s perceived to create revenue and profit (and yes, skills are a major factor in that).

I’d certainly not say it’s everything, look at all the highly-paid mediocre CEOs. Education has rigorously been shown to lead to higher incomes and wealth on average.

Yup! I was a part of the learn to code industry. I am proud of that, bc I know my worker helped a lot of marginalized people gain wealth and power (woo!). My own occupation, stats and econometrics, requires years of higher education to even begin (and decades to master), and yet ~ half of SWE were looking down on me, disrespecting me. To be clear, there were many who were not, but usually they were from some marginalized group: women, autistic, person of color, gay, person from a poor country, etc. I thought, why is my towering knowledge not being respected? Ah, the patriarchy combined with SWE. And then as time went on I just started using my knowledge for myself / those that know and that’s worked out well (bc it’s based on actually knowing math as opposed to relying on the patriarchy).

I think it’s possible the industry eventually figures out that statisticians and econometricians know far more than CS / SWEs (bc AI will tell people), but it could be a decade from now.

Agree, AIs are better decision-makers on average than people (just look at the grifters we've given power to). These are machines that can perform more advanced mathematics than even the most advanced mathematicians.

Sadly I think this post will mislead people, bc the difficult truth (for many) is that software engineering isn't that hard and that's why AI can easily substitute that layer (lower barriers to entry than widely believed).

I don’t think there are other models near Fable’s capabilities.

Ah thank you for updating me there.

In what ways? LM Arena has Opus 4.7 w/ 1567 -/+ 7 vs. 1505 -/+ 10 from GPT-5.5 Codex in code. I'm currently using both.

Admittedly my recent experience tilts Opus now 4.8, but you and others have my interest piqued re: GPT-5.5 Codex so I'm trying that more now.